Difference in voice analysis result by pre- and post- processing of telephone line

Area of Science:

• Psychiatry and Psychology
• Speech and Audio Processing
• Biomedical Engineering

Background:

• Voice analysis is a promising, non-invasive method for assessing mental health conditions like depression.
• Estimating depressive tendency, termed Vitality, requires accurate voice data.
• The feasibility of collecting voice data remotely via telephone lines for such analyses remains a key question.

Purpose of the Study:

• To investigate the impact of sound quality deterioration from telephone line transmission on voice analysis for Vitality estimation.
• To determine if telephone-recorded voices are suitable for assessing depressive tendencies using MIMOSYS software.

Main Methods:

• Prepared voice samples from approximately 1,000 individuals, initially recorded with a high-quality recorder.
• Applied coding and resampling to simulate voice transmission over standard telephone lines.
• Compared Vitality scores derived from voice analysis (using MIMOSYS) before and after simulated telephone transmission.

Main Results:

• A high correlation was observed between Vitality scores obtained before and after the voice samples underwent coding and resampling for telephone transmission.
• The results indicate minimal detrimental impact of telephone line sound quality degradation on Vitality estimation.
• MIMOSYS analysis demonstrated robust performance despite simulated telephone line interference.

Conclusions:

• Collecting voice data via telephone lines is a viable and effective method for Vitality estimation and depressive tendency assessment.
• This finding supports the use of telecommunication technologies for remote mental health monitoring through voice analysis.
• The study validates the utility of MIMOSYS for analyzing voice data transmitted over potentially degraded phone line channels.